In most SMEs, the complaints process doesn't exist as a process. It exists as one person's inbox, a file called "RMA-2024-v3-FINAL.xlsx" and decisions made "by feel". Until the customer calls for the fifth time asking about their return — and nobody in the company can answer.
This isn't a problem of bad people or a bad company. It's a problem of a process that grew on its own — because nobody ever sat down and designed it. In the early years there were five complaints per quarter and a salesperson handled them between coffees. Then sales grew, products got more complicated, and the process stayed the same: "if something comes back, we'll deal with it". Once "something comes back" 30 times a month, you have a drama on your hands.
RMA (Return Merchandise Authorization) sounds like corporate jargon, but it describes something very simple: the path goods and money take from the moment the customer says "doesn't work" or "I want to return this", to the moment the case is closed. In a well-run company, that path has a number, an owner at each stage, and a single source of truth — usually the ERP.
Why Excel stops working
Excel is brilliant for keeping a list of 30 items that one person looks at. RMA is usually a process that involves the warehouse (receiving the goods), service (diagnosis), accounting (issuing a corrective invoice), the salesperson (contact with the customer), and sometimes the boss (decision on more expensive cases). Excel can't handle five people at once, doesn't chase statuses, doesn't count days since submission and doesn't remind anyone of anything.
The result is always the same. At one of our clients — a 24-person consumer electronics distributor — an RMA audit after six months found: 47 cases "open" for more than 60 days, including 12 that everyone had simply forgotten about. The losses on that "forgotten pile" came to around 38,000 PLN — partly goods whose manufacturer warranty had long expired, partly shipping costs nobody had billed anyone for, partly customers who let it go but didn't forget. Excel didn't fool anyone. It just wasn't the right tool for the job.
What a well-organised process looks like
It starts with one thing: every case gets a unique number — ideally from a counter in the ERP, not from a salesperson's head. The RMA number is the common language of the whole company. Without it, the warehouse says "that box from the lady in Kraków", accounting says "that corrective invoice from last Thursday", and the boss says "the one from the printer complaint". Nobody realises it's a single case.
Then you need a very short list of statuses — not 15, but 5 or 6 that reflect reality. A tried-and-tested set for SMEs: submitted → accepted → goods received → in diagnosis → decision → closed. Every status has a clear owner: who creates it, who changes it, who closes it. If a case sits in "in diagnosis" for a third week, the ERP has to notice and remind someone — without that, we're back to the forgotten-pile Excel problem.
The third thing is the decision made in advance, not on the side. In most companies the salesperson decides on the fly whether to accept a complaint, replace the goods or refund the money. That works as long as all salespeople decide the same way — which is never. Instead, keep three simple rules: up to what amount the salesperson decides alone, above what — the manager, and when the case goes to the manufacturer's service. Rules on one page, known to everyone, wired into the ERP as picklist fields.
What the ERP has to do
You don't need a dedicated "RMA module for 40 grand". Most ERPs used in Polish SMEs — Comarch Optima, enova365, Subiekt GT, Microsoft Dynamics — have this built in or need only a small extension. Specifically, four things.
- Link the case to the sales invoice and the specific line item. Without that, you don't know when it was bought, at what price, whether the manufacturer's warranty is still valid. Retyping the tax ID and invoice number into Excel guarantees a typo.
- Auto-generate the goods receipt (return GRN) once accepted. The warehouse has to know what to expect before the box arrives — otherwise it lands "on the pile" and stays there.
- One-click corrective invoice when the case is closed with "refund". Manually writing corrective invoices is the main source of accounting mistakes — and an excuse for accounting not to close the case for another week.
- A "my cases" view for every participant. The salesperson sees theirs, the warehouse theirs, the boss everything. Without that, everyone looks at everything, and nobody owns any specific case.
This isn't a year-long wishlist. This is the minimum below which the process doesn't work. If your ERP doesn't have this, it doesn't mean you have to replace it — more often, a small in-house API and a lightweight panel next to it will do. But without these four things, you stay in Excel, only more expensive.
Metrics worth tracking
A process without measurement reverts to chaos — quietly, over about six months. Four numbers are enough to see whether RMA is working:
- Average time to close (from submission to decision). Healthy for an SME is 7–14 days depending on the industry. If yours is 45, something is off.
- Share of cases older than 30 days. Should be under 10%. Above 20% — the process urgently needs unblocking.
- Accepted vs rejected. If you accept 95%, salespeople are probably taking the easy route. If 30%, you're probably making enemies of your customers.
- RMA cost as a share of revenue. For electronics distribution, 2–4% is normal noise; above 6% something needs fixing — in the product, the website descriptions, or with a specific supplier.
Back to that distributor: three months after rolling out the process (statuses, invoice link, automatic return GRN, weekly boss report), the average time to close fell from 28 to 9 days, and the "forgotten pile" disappeared — not because someone suddenly solved everything, but because the system reminded people daily until they closed the cases. Separately, and measurably: customer calls asking "what's happening with my return" dropped by two-thirds.
Traps that are easy to fall into
A few things we regularly see in these rollouts — worth avoiding from day one:
- Too many statuses. Someone proposes 12, "because every situation is different". A month later half the fields are empty and the other half wrong. 5–6 statuses is the maximum people can handle in daily work.
- A separate RMA system alongside the ERP. Tempting, because vendors sell dedicated tools. Problem: data about the goods, customer and invoice already lives in the ERP, and moving it back and forth is a written invitation to inconsistency. Keep RMA where the invoice already lives.
- No internal SLA. If there's no rule saying "diagnosis in 3 working days", nobody does it in 3 days. It's not about punishing people; it's about them knowing when the case passes to the next person.
- Closing a case "for later". A case closed without the corrective invoice isn't closed — it's an accounting bomb waiting at year-end. Automatic corrections from the ERP kill this sin at the source.
Where to start
Not with buying a tool. With half a day, a piece of paper and five people from the company: the salesperson, the warehouse, service, accounting and the boss. Draw the current process together — from memory, without prettifying it. You'll see that in three places nobody knows who should be doing what. Those are the points where RMA is bleeding your money.
Only then configure it in the ERP, a simple panel for the salesperson, one weekly report for the boss. No revolution, no 40-grand module, no year-long project.
Complaints don't disappear if you ignore them — they just turn into a quiet cost nobody sees in the P&L.